You already know the feedback before the tutor says it. The board looks great, the images are beautiful, someone will ask why the Therme Vals photo is there, and you'll say something about materiality and atmosphere and the room will go quiet. Not because you're wrong. Because you've brought a mood and they asked for evidence.
The fix isn't collecting better images. It's changing the order of operations. Precedent research that changes a design starts from a problem you already have, not from a project you already like. Everything else in this article follows from that one reversal.
Here's the difference, in the only terms that matter in a crit.
A mood board says: I would like my building to feel like this. Nobody can argue with it, which is exactly why it carries no weight. It cannot be tested, so it cannot be defended.
A precedent says: this building had my problem, here is what it did, here is the measurement, and here is where it fell over. That can be argued with. That's the point. A claim that can be attacked is a claim that can be defended, and defending it in front of a panel is what makes a scheme look designed rather than assembled.
What we've found teaching this: students who can't explain their precedents usually chose them last. The design came first, the images were recruited afterwards to make it look researched, and the tutor can smell it from across the room. Precedent chosen after the fact is decoration by definition, no matter how rigorous it looks.
Start From the Thing That's Broken
Before you open a single tab, write down the design problem in one sentence. Not the brief. The stuck bit.
"The north-facing studios get no useful daylight." "I have 34 flats sharing one core and the corridor is grim." "The section has a 4m level change I keep drawing a ramp through and it eats half the plan." "The public route and the private route cross at the front door and it's a mess."
That sentence is your search. Not "timber housing" or "brutalist library" — those are aesthetics, and searching an aesthetic gets you an aesthetic. Searching a problem gets you a strategy.
This reframing does something useful and slightly uncomfortable: it opens up precedents you'd never have pinned. A 1960s hospital might solve your corridor problem better than any of the Scandinavian housing you've been hoarding. A car park might solve your level change. If the problem is the search term, ugly buildings become useful, and a lot of the beautiful ones turn out to have nothing to say to you.
As a rule of thumb we use: if you can't name the problem a precedent solves for you, it isn't a precedent, it's wallpaper. Delete it or move it to a separate folder called things I like. Keeping that folder is fine. Just don't put it on the board.
Choose for the Problem, Not the Photograph
Once the problem is written down, you're filtering on things a photograph can't show you.
Brief similarity. Does it hold roughly the same programme at roughly the same scale? A 300-pupil primary school and a 1,200-pupil academy are not the same building with different numbers; the circulation logic breaks completely somewhere in between.
Site and climate. A courtyard plan that works in Seville is a wind tunnel in Aberdeen. If you're citing a building from a different latitude, you need to say what you're taking — the geometry, not the performance — and what you're changing.
Constraint. Tight urban plot, awkward level change, listed neighbour, ferocious overlooking, a budget that couldn't stretch past blockwork. Buildings that were squeezed are more instructive than buildings that were free, because you can see the decision in the plan.
Age. A building that has been occupied for twenty years has a service record. A building photographed six weeks after handover has a photographer.
The Four Extractions
For every precedent that survives the filter, get four things out of it. Four, not forty. The discipline of stopping at four is what makes this doable for six precedents rather than heroic for one.
1. The plan strategy, in one diagram
Redraw the plan. By hand, small, at the same scale as every other precedent you're studying. Strip out everything except the move: where does the public enter, where does it stop, where does the servicing go, what's the repeated unit, what's the exception.
You are looking for the parti — the single organising idea the whole plan obeys. Bar with a comb of rooms. Doughnut round a void. Two blocks and a joint. Nine-square with the middle taken out. If you can't reduce it to something you could draw in five seconds on a napkin, you haven't finished looking.
Same scale is not a detail. Same scale is the whole trick. Half of what students think are formal preferences turn out to be scale differences they never noticed, and you only see it when the two plans sit on top of each other at 1:500.
2. One measured dimension
Take one number off the drawing and write it down with units. The corridor width. The floor-to-soffit height in the space that does the work. The structural grid. The depth from window to back wall. The rise of the stair that makes the section function.
This is the single highest-return habit in the whole method, and almost nobody does it. A dimension is the thing you can test against your own plan tonight. If the precedent's gallery is 6m deep and yours is 11m, you've just learned why yours feels like a corridor, and you learned it in thirty seconds rather than at the crit.
It also changes how you speak. "The circulation is generous" is a taste claim. "They ran it at 2.4m where mine is 1.5m, and that's why theirs takes furniture" is an argument. In my experience that one shift — from adjective to dimension — does more for a student's crit performance than any amount of extra rendering.
3. The section move
Most precedents are copied in plan and lost in section, which is odd, because the section is usually where the idea lives. Split level. Stacked floor heights that aren't equal. A double-height space put deliberately where the plan is deepest so light can get to the middle. Roof pitched one way to keep the north light and dump the rain somewhere useful.
Draw one section through the precedent yourself, even a rough one from photographs and a floor plan, and mark the one move that makes it work. If a project has no section move — if every floor is the same and the interest is all in the elevation — say so. That's a finding. It tells you the building is solving a different problem to yours.
4. The failure
This is the one that separates research from fandom, and it's the one tutors notice instantly.
Every building has a compromise in it. Find it. The signs are cheap to spot once you're looking: blinds permanently drawn on the glazed façade, a fire door propped open because the route doesn't work, laminated A4 signs telling people where reception actually is, a beautiful entrance nobody uses because everyone comes from the car park at the back, a courtyard that's empty in every photograph taken after the opening.
How to find it: visit if you can, read the critical reviews rather than the press release, look at recent photographs rather than the handover set, and talk to whoever is on the front desk. Building users will tell you the truth about their building within ninety seconds of being asked, and nobody ever asks them.
Naming the failure and designing around it is precedent use at its most convincing. "I've taken their courtyard geometry but pulled the glazing back a metre under the overhang, because in theirs the south rooms overheat and the blinds are down in every photo I could find" is a sentence that ends a line of questioning. It shows you looked at the building rather than the picture of it.
Stop Scraping, Start Sourcing
Pinterest and ArchDaily are discovery tools, not research. They're great for finding out that a building exists. They are close to useless for the four extractions, because they publish photographs and — if you're lucky — a plan with no scale bar.
The things that actually carry the information: published monographs, technical journals that print details rather than money shots, drawings released through planning portals for anything recent and local, university and institutional archives, and the building itself. A site visit to a mediocre local precedent beats a scraped folder of a famous one every single time, because you can measure the corridor with your feet and photograph what's gone wrong.
On quantity: three or four precedents interrogated properly will carry a scheme. Thirty collected will carry nothing. What's convincing in a crit is depth and the comparison between them — the same four questions asked of each, so the differences mean something — not the size of the pile.
Write It Down Like You'll Have to Defend It
Keep a record for each precedent with the same fields every time: project, architect, date, location, the problem it solves for you, the parti diagram, the one dimension, the section move, the failure, and where you got each piece of information from.
That last field does two jobs. In a crit it means you can answer "where did you get that?" without flailing. And if you end up in practice, that same record is what turns into a paragraph in a design and access statement — in practice, precedent cited with a source, a dimension and a reason is a great deal harder for a planning officer to wave away than a page of photographs of nice brick buildings.
Do it as you go. Doing it at the end is how the whole thing collapses back into decoration.
What the Tutor Is Actually Listening For
They're listening for one thing: did the precedent change a decision, and can you point at the decision.
So point at it. "I moved the core off-centre because of this." "I dropped the ceiling to 2.4m at the entrance because of this, and raised it in the hall for the same reason they did." "I rejected this one — same programme, but their plan depends on a service yard I haven't got."
That last move — the rejected precedent — is worth more than three admiring ones. Showing a project you studied properly and then declined to follow proves the studying was real. Nobody rejects a building they only ever looked at.
The images can stay on the board. Keep them. Just make sure that under every one of them there's a diagram you drew, a dimension you took, and a sentence about what it made you do differently. That's the whole difference between a board that looks researched and a scheme that is.
